COMMENTARY | “Aren’t you going to enter the campus March Madness Tournament, Dr. Tures?” my students asked me. “You’re always talking about how you can use statistics to predict future events in politics.”
It’s true that I can be a bit of a Nate Silver type when it comes to elections, international conflict, and even terrorism, using numbers to do that. So in the five minutes before class, I whipped out a pen and filled out a bracket that a student gave me.
Just before the Final Four, the Dean of Students handed me my prize. “But the tournament’s not over!” I exclaimed. “You’ve mathematically eliminated everyone else,” he replied.
Everyone’s got a story about picking the NCAA Tournament. I remember getting UCLA, Kentucky and Arkansas right in the 1990s. But that tournament, where Duke edged Butler, was my best.
How did I fill it out? I looked up the RPI (Rating Percentage Index) and picked from the top down. I’ve also had luck comparing RPI from the bottom-up (matching team by team each game, and working upwards). But over the last two years, I’ve come up a little short. Louisville was #2 in the RPI last year and won, but Michigan was 11th in RPI. Syracuse was tenth and Wichita State was 18th. I was good, but not good enough.
Some go by record. Doing that would have netted you Gonzaga (top 12), Louisville (number 2), Memphis (18th in the end) and Kansas (elite eight). But you would have whiffed big time on Wichita State (24th) and Syracuse (30th), both of whom made the Final Four. Middle Tennessee State had the seventh best record, but was nowhere near winning it all.
Some swear by strength of schedule. that would have helped you with Louisville (#3 SOS) and Syracuse (#4) but you would have guessed badly on Minnesota (#5), North Carolina (#9) and New Mexico (#7), all of whom were early casualties. Michigan was #14 in Strength of Schedule and Wichita State was 39th in SOS, but both made the Final Four.
Think the USA Today Coaches’ Poll was any better? Think again. Louisville was #2 in that and won it all, but Gonzaga, Kansas State etc weren’t Final Four material. With Final Four attendees Michigan (#11) and Syracuse (#18) and Wichita State (unranked…and not receiving any votes) well behind, you might as well have gone by one of those formulas picking by color, dogs vs. cats, etc. The Associated Press did okay, but could have done better.
I thought about defensive efficiency, a trendy idea. That 2012-2013 ranking had Louisville and Syracuse near the top. But Wichita State was in the forties and Michigan was ranked #97 in defense efficiency. Clearly the Wolverines and Shockers were there for their scoring.
To determine the best, I looked at correlation coefficients for winning percentage, RPI, strength of schedule, and Associated Press Ranking, all found here. RPI had the strongest correlation with the final outcome, followed by AP ranking, then winning percentage. Strength of schedule finished the worst.
So look for Arizona and Florida in the final game, according to the 2013-2014 RPI rankings. Kansas and Villanova should make the Final Four. Other “Elite Eight” teams will include Wichita State, Wisconsin, Creighton and Duke.